Catalyst Decomposition during Olefin Metathesis Yields Isomerization‐Active Ruthenium Nanoparticles

The second‐generation Grubbs catalyst, RuCl2(H2IMes)(PCy3) (=CHPh) [GII; H2IMes=1,3‐bis(2,4,6‐trimethylphenyl)‐4,5‐dihydroimidazol‐2‐ylidene, Cy=cyclohexyl], is shown to decompose during olefin metathesis to generate Ru nanoparticles (RuNPs). These RuNPs appear to contribute significantly to competing isomerization during metathesis. Larger, partially oxidized RuNPs are also observed in commercial GII, but these exhibit modest isomerization activity. Removal of RuNPs from the precatalyst does not prevent isomerization, because new, more reactive NPs are generated by catalyst decomposition during metathesis.

Ruthenium-catalyzed olefin metathesis is ac ore tool in organic synthesis [1] anda ne merging protagonist in the pharmaceutical industry. [2] Notwithstanding the importance of these advances, an umber of reports cite challenges arising from competing olefin isomerization, [3] the dominant non-metathetical side reaction. [4] Isomerization is particularly pronounced for the second-generation Grubbs catalyst (GII), relative to its predecessor GI (Figure 1). [3] Ta ndem metathesis-isomerization or isomerization-metathesis protocols, employed as ad eliberates ynthetic strategy,c an enable access to targets that are otherwise challenging or inaccessible. [5][6][7] More commonly,however, isomerizationisanunintended, often capricious side reactiont hat results in variable control over product selectivity and yields, in processes rang-ing from ring-closing metathesis (RCM)t oc ross-metathesis (CM) and metathesis polymerization. [2,3,8,9] Rutheniumh ydride complexes generated by catalyst decomposition are widely viewed as responsible.U ntil now,o nly molecular complexes have been considered as potential culprits, despite the low isomerization activity documented for leading candidates. [10] Herein,w es how that ruthenium nanoparticles (RuNPs)a re formed by decompositiono fGII duringm etathesis, and that these are important,h itherto unrecognized contributors to competing olefin isomerization.N otably,w hereas NP formation is common for low-coordinateP dcatalysts that cycle between Pd II and Pd 0 , [11] reports of such behaviorfor well-defined monoruthenium complexes operating in organic mediaa re rare, outside hydrogenation reactionsm ediated by h 6 -arene complexes of ruthenium. [12] This is the first report of metal NP formation by decomposition of am olecular metathesis catalyst.
Olefin isomerization by RuNPs has not, to our knowledge, previously been reported. Given the activity of such entitiesi n other catalytic contexts, however, [13] we speculated that they might functiona sv iable isomerizationc atalysts. This provedt o be the case. RuNPs werep repared by ar ange of methods (see the Supporting Information) [14][15][16] and were tested for their activity toward isomerizationo fe stragole (1). Estragole is an important renewable allylbenzene used in metathesis reactions, [17] which, asw ith its congeners, [9,18] is readily isomerized. Figure 2 shows the isomerization activity recorded for four different Rucontaining nanostructures.A ll are clearly capable of inducing 1!2 isomerization. By far most active, however,w ere the Chaudret-Philippot NPs (type D), prepared under rigorously anaerobic conditions, ands tabilized by N-heterocyclic carbene (NHC)l igands. [14,19] The dramatically higheri somerization activity of these NHC-stabilizedN Ps is consistent with the absence of oxidized surface species.
Given this evidence that RuNPs promote olefin isomerization, and prior reportst hat such side reactions declined if com-    mercial GII was chromatographed prior to use, [20,21] we questioned whether RuNP contaminants might be present in GII, [22] which trigger competing isomerization during metathesis. We found that commercial GII catalysts do indeed contain RuNPs, present as aggregates that agglomerate on isolation to an average size of > 500 nm (see the Supporting Information). However,t he isolated particles induced olefin isomerization with low efficiency.T hey required 24 ht or each 45 %y ield of 2 under the conditions of Figure 2. This is unsurprising given their large size and partial oxidation, both of which limit the number of active surface sites.
To determine whether isomerization could be inhibited by removing the RuNPs present in the precatalyst, we generated NP-free GII by ultracentrifugation under an atmosphere of N 2 . As illustrated in Figure 3, the purified GII effected both metathesis and isomerization of estragole (1). Thus, yields of metathesis product 3 increased over the first houro ft he reaction but then declined as 3 underwent isomerization (Figure 3a). Strikingly,t he extent of isomerization was only 15 %l ess than that effected by non-purified GII (Figure3b). Freshlyd ecomposed Ru products thusa ppear to be important contributors to isomerization,w ith al evel of activity much higher than that of the RuNP impuritiesp resent in the precatalyst.
Also notable in Figure 3b is the approximately 30 mininduction period that precedes the onset of isomerization. Formation of NPs over this timescale was confirmed by in situ nephelometry experiments,i nw hich the intensity of scattered light was detected by synchronousw avelength scanning. As with conventional dynamic light scattering, increases in scattering intensity indicate NP formation. Intensity changes were monitored in the l = 600-700nmr egion to eliminate perturbation arisingf rom absorption by the sample. The intensity of scattering increased over the first 30 min (see the Supporting Information), ac hange that maps onto the induction period in isomerization.I nt he absence of substrate, scatteringw as significantly reduced.
This evidencei mplies that RuNPs are formed by decomposition of ruthenium species generated during metathesis. We attributet he formation of nanoparticles, as opposed to molecular Ru products,t ot he loss of multiple ligands in the process of catalystdecomposition.R elevant in this contextist he established pathway by which free PCy 3 ,l iberated from the resting-state complex GIIm (Scheme 1), attacks the methylidene ligand of active species Ru-1. [23,24] Eliminationo ft he s-alkyl ligand thus formed occurs by abstractiono faproton (most plausibly from the H 2 IMes ligand) and boundc hloride. This process culminatesi nt he extrusiono f[ MePCy 3 ]Cl (A), an et loss of three ligandsp er Ru center.W hereas isolation of the putative s-alkyl intermediate Ru-2 is precluded by its short lifetime, we recently succeeded in trapping out such ac omplex in the first-generation Grubbs system. [25] The detailso fN Pf ormation are now being probedb yi nsitu X-ray absorption studies, but the lowcoordinate Ru species resulting from such "ligand stripping" represents aplausible startingp oint.
Further experimental evidencef or RuNP formation during metathesis comes from electron microscopy.I nt hese experiments,s tyrene 4 wasc hosen as the substrate, because the low solubility of its self-metathesis product 5 facilitates removal of organic speciest hat otherwise occludet he micrographs.S canning electron microscopy (SEM,F igure 4a)r evealed NP-free solutions. Likewise, transmission electron microscopy (TEM) showedn oN Ps in analysis of multiple samples, down to the 0.2 nm detection level of the instrument. In contrast, abundant RuNP formation was evident following metathesis of 4,a s shown in Figure 4b.
To examinew hether isomerization is promoted by RuNPs generated by catalyst decomposition during metathesis, or by molecular species formed at an earliers tage, we performed  www.chemcatchem.org mercury-poisoning experiments.P oisoning by elementalm ercury is ac ommon test for the involvement of surface-active metal(0) sites in catalysis. [26][27][28] As shown in Figure 5, isomerization of 1 droppedb ya pproximately 50 %i nt he presence of Hg. Control experiments indicated that Hg had an egligible impact on the isomerization activity of common Ru hydride complexes (see the Supporting Information), or on the formation of A.I ndeed, the Hg test may under-report the contribution of RuNPs in Figure 5, given the reported instability of the Ru-Hg amalgam [29] or adsorbate. [26] Substoichiometric poisoning experiments ( Figure 6) were performed to further probe the involvement of RuNPs in isomerization.S uch experiments are predicated on the requirement for ! 1equivalent of ap oisoning ligand to inhibit catalysis by molecular Ru species, in contrast with the smaller number of ligands required to inhibitN Pc atalysis (in which much of the initial metal charge is inaccessible in the NP core). Accordingly,w ea ssessed the impacto fP Me 3 ,P (OMe) 3 ,a nd PPh 2 Me (0.1 equiv.v s. GII)o nt he rate of isomerization during self-metathesis of estragole (1). These experiments were performed at 24 8Ct om aximize the poisoning effect. [30] To compensatef or the negative impact of the lower temperature on catalysis, we used ab atch of estragole that showed much higher rates of isomerization. [31] Isomerization ceased immediately on adding the phosphine/phosphite poison ( Figure 6).
The foregoingd emonstrates that RuNPs can show high activity for olefin isomerization, that RuNPs are formed by catalyst decomposition during GII-catalyzed metathesis, that Hg poisoning reduces isomerization, and that the addition of as mall proportion of ap hosphine or phosphite poison, relative to the total Ru loading, is sufficient to completely shut down isomerization. On the basis of this cumulativep icture, we proposet hat RuNPs formed by catalystd ecomposition are important contributors to unwanted isomerization during olefin metathesis.
The contexta bove focuses on unintended isomerization as ap roblem encountered duringo lefin metathesis. Insight into its origin, however,p oints toward new opportunities. Ther eaction conditions explored above wered esigned for metathesis, rather than nanoparticle formation or isomerization.O ptimizing the synthesis of ruthenium nanoparticles, as well as the isomerization conditions, is expected to open new doors for the design of novel isomerization catalysts. Keywords: catalyst decomposition · isomerization · metathesis · nanoparticles · side reactions